改进后的轮换步长PSO算法

Improved Rotation Step PSO Algorithm

轮换步长PSO算法是针对PSO算法的早熟问题提出的一种新型PSO优化方法。此算法在粒子之间加入排斥力,把即将陷入早熟的粒子分开,使得各个粒子无法聚集在同一点上,从而使整个粒子群不会发生全部粒子收敛于局部最优解的现象,以提高算法的抗早熟性。轮换步长PSO算法虽具有很好的抗早熟性,但其步长在方向上不具有随机性,而烟花算法在方向上具有很好的随机性。因此,将轮换步长PSO算法和烟花算法相结合,大大改善了轮换步长PSO算法在步长方向的随机性。

Aiming at resolving the premature problem of PSO,a new optimization method was proposed,which is called rotation step PSO algorithm.This algorithm adds repelling forces between particles to avoid them gathering at the same point.Consequently,it is impossible for the entire particle swarm to converge in local optimal solution.Thus,the anti-precocity character of PSO is improved.Although the rotation step PSO is good at the anti-precocity aspect,its step is not random in direction.On the contrary,fireworks algorithm has fine randomness in direction.In this paper,rotation step PSO and fireworks algorithm is combined,so the randomness of the former is effectively improved.